- Added a standalone TTS module - Introduced a new TTS generation module in the subtitle editor for easier speech synthesis from translated text - Added support for the WhisperMLX transcription engine (macOS 14.0+) - Added WhisperKit Diarization (speaker separation) engine - Updated FluidAudio engine to V0.13.7 - Updated WhisperKit engine to V0.17.0 - Real-time transcription now supports direct output to OBS-compatible files - Added several tools for batch cleaning special characters in subtitle lines - Directory monitoring now supports linking to specific project groups - The real-time transcription floating window can now be displayed and moved across fullscreen app spaces - Fixed several issues with the preset feature - Fixed an issue where exporting video segments could fail

- Updated the whisper.cpp engine to v1.8.3, with Silero VAD enabled by default - Updated the WhisperKit engine to v0.15.0 - Updated the FluidAudio engine to v0.12.0 - The minimum supported system version has been raised to macOS 13.3+ - Subtitle line thumbnails can now be generated automatically when the speaker changes - Subtitle line thumbnails now support exporting screenshots at the original video resolution

- Added support for Korean, Japanese, and Cantonese as secondary languages when using the Apple transcription engine for real-time transcription. - Fixed several issues when switching the quick player to fullscreen mode - Fixed an issue where the translation option could not be enabled when using the Apple engine for real-time transcription

- The project list sidebar now supports direct video preview - Fixed UI issues related to the translation component - Updated FluidAudio to v0.10.0

- Refactored the control logic for VAD and noise reduction options - Refactored the processing workflow for multi-track audio files - Upgraded FluidAudio to v0.8 - Fixed an issue where exporting MP3 files could fail - Fixed timestamp issues when using WhisperKit in VAD mode - Fixed other crash issues

- Added support for translating subtitle lines using the new built-in translation engine on macOS 26 - Improved various details related to real-time transcription - Restored the model parameter settings for whisper.cpp and WhisperKit in the execution panel - The project sidebar can now directly open the floating real-time transcription window - Updated the FluidAudio transcription engine to version V0.7.10 - Fixed an issue where the floating transcription window could not be launched from the system status bar - Real-time audio-only transcription now also supports Apple’s built-in transcription engine - Fixed an occasional issue where double-clicking an item in the table failed to open the project - Fixed several UI issues observed on macOS 26

- Added one-click export to Obsidian - Added support for adding folders to the project list, which are automatically converted into project groups for quick batch project addition - Added option to export audio and subtitles as an MP4 file containing only audio tracks - Added post-transcription automation to automatically perform speaker diarization - Improved the speaker replacement and common speaker settings control bar - Subtitles for different speakers now appear in distinct colors after speaker recognition - Added an option in auto-export to include the extracted WAV audio file from the media during transcription - Upgraded FluidAudio transcription engine to version V0.7.8 - Optimized handling when dragging projects into sidebar groups - Fixed an issue where Korean text spacing appeared incorrectly when using Apple’s built-in AI transcription model - Various minor bug fixes and performance improvements

- Improved the interface for re-transcribing transcription segments, making it easier to compare the new results. - Added Kokoro voice generation, which can create speech files based on subtitles (currently available for English only). - Enhanced the directory monitoring feature to also watch for new files added in subdirectories and automatically trigger transcription, including files synced from iCloud on mobile devices. - Upgraded the FluidAudio transcription engine to V0.7.7. - Fixed an issue where the video previewer produced distorted sound when first opened. - Fixed an issue where the first frame of exported videos with subtitles could skip frames.

- Automated actions now support executing predefined AI operations automatically when a project is completed - Automated AI operations now support using Apple’s built-in large language model for project summarization - Segment re-transcription now supports the FluidAudio engine - Fixed an issue where selected segments could not be re-transcribed multiple times - Fixed a timestamp mismatch issue when re-transcribing with Apple’s built-in transcription engine - Removed several plugins that can no longer be adapted - Other minor bug fixes

- Real-time transcription can be started directly from the main interface - The auto-export feature now supports exporting in the original file’s directory - Added a QuickLook plugin to preview the content of SRT files - SRT files can now be translated directly - Fixed an issue where the end of sentences in real-time transcription was duplicated - Fixed several bugs in the auto-export feature - Fixed some UI anomalies on macOS 26 - Fixed an issue where the Apple built-in AI transcription model did not appear in the transcription engine selection list

- Added support for using Apple’s built-in AI transcription model on macOS 26+ for transcription. - Changed model parameter settings to a sidebar format (can be reverted to the old popup style in settings). - Automatic export now supports selecting multiple templates simultaneously. - Fixed some UI issues on macOS 26.

- On macOS 15 and later, real-time recording now supports capturing and transcribing microphone audio simultaneously - Real-time transcription in audio-only mode no longer requires ScreenKit permission (macOS 14+) - Fixed an issue where resources were not properly released after closing the real-time transcription window - Upgraded FluidAudio transcription engine to version 0.6.1 - UI adapted for macOS 26 - Other minor fixes and improvements

- Added project group sidebar to organize projects into different groups - Added floating video subtitle preview mode - Redesigned the application settings interface - Updated whisper.cpp transcription engine to v1.7.6 - Updated whisperkit transcription engine to v0.13.1 - Updated fluidaudio transcription engine with support for parakeet-tdt-0.6b-v3 and timestamp display - Deepgram transcription engine now supports the latest nova-3 model - Subtitle editor can now display transcription content only (useful for projects that don’t require timestamps or markers) - Subtitle previewer now automatically scales subtitle font size based on video resolution - Subtitle previewer now supports customizable line spacing - Fixed an issue where automation actions were not executed when using fluidaudio and whisperkit engines - Simplified translation component UI - Fixed several UI issues on macOS 12 and 13 - Other minor fixes

- Added support for using the parakeet-tdt-0.6b-v2 model in the transcription engine - When using the Whisper engine, VAD can now be enabled to automatically split audio and reduce hallucinations - Fixed several issues with the preset feature - Optimized button styles in dark mode - Fixed an issue where some videos could not be displayed properly in the quick cut window - Fixed an issue that prevented exporting only the selected rows when exporting videos - Fixed performance lag in the Video Export tab - Bug fixes and performance improvements

- Added the ability to save the current project’s transcription model settings as a preset for reuse in new projects - Automatic transcription via directory monitoring now supports multiple directories - Translation and LLM tools now support the local Ollama protocol - Improve directory monitoring to better detect the download status of files in iCloud directories, with clear logs showing the download activity of cloud files. - Simplified the usage of plain text mode - Fixed performance issues in LLM chat mode when handling large data volumes - Added an option to switch back to PyAnnote for diarization integration - Real-time transcription now supports using GPT-based translation engines - Directory monitoring now automatically triggers transcription without needing “auto-start” enabled

- Optimized the LLM configuration interface and debug logs for clearer presentation and easier debugging - Added support for increasing or decreasing the audio volume during video export - Introduced new post-processing tools in the Tools panel; for example, when using the large-v2-dv1a-diarization model, subtitles can now be reordered based on detected speaker labels - Fixed an issue where the subtitle preview was not refreshed automatically after executing LLM or post-processing tools - Fixed a problem where the ESC key did not work in the real-time transcription window - Fixed several issues encountered during first-time use - Minor bug fixes and improvements

- Optimized the property configuration interface for SRT and ASS subtitle formats, adding more parameters to control subtitle styles. - Enhanced the video export interface with support for exporting videos at reduced resolutions. - Improved the method of merging and splitting subtitles based on common punctuation marks. - Added a post-processing tool in the quick tools panel to automatically wrap subtitle lines according to length. - Enabled direct import of FCPXML subtitle files into the subtitle editor for editing. - Added an option to burn videos or GIFs without subtitles during hardcoding. - Fixed an issue where the window would behave abnormally when restoring from fullscreen to original size. - Fixed the problem where preview video size was not recorded after closing the project. - Fixed jittering issues when editing line breaks in the quick subtitle editor.

- Added a new diarization method for improved speaker separation. - Integrated large language model (LLM) features into the new Inspector panel with simplified configuration options. - LLM now supports translating the entire SRT subtitle file or returning a revised SRT version for direct replacement in the subtitle editor. - Video export now supports exporting segments as GIF animations. - Improved the layout of the subtitle editor. - Added quick-edit functionality for subtitle text, available under the bottom tab of the Inspector on the right. - GPT-based translation and LLM features are now compatible with Gemini, DeepSeek, and OpenRouter APIs. - Added batch extraction of video thumbnails at subtitle start times, displayed within the subtitle editor. - Added keyframe detection for scene changes, with markers shown on the waveform editor.

- Added a visualization feature based on audio energy bars, allowing users to drag and adjust the start and end times of subtitles (for dialogues and audio without background music, an automatic alignment feature is supported to adjust subtitle timing automatically). - Added a line erase function; erased subtitle lines will be shown as deleted and will not appear in subtitle previews or exported files. - Fixed an issue where some projects could not start batch execution after being selected in bulk. - When saving real-time transcription files, the snapshot name is automatically used as the filename. - Added a quick access button in the project list to directly enter text mode. - Fixed an issue where text mode during project transcription would not automatically scroll to the last line. - Added a post-processing text feature to split lines that contain multiple “-” bullet points into separate lines. - Optimized export and media cutting functions: if the selected subtitle lines are not continuous, the system will skip unselected lines during the cutting operation.

- Added feature to add chapters to videos. Existing chapter information can be automatically imported, edited, and re-exported with the video file - Added support for automatically importing embedded text-based subtitles from MP4 files as transcription results. This allows you to quickly translate existing subtitles in a video into other languages - Added a new free translation engine - Upgraded the WhisperKit transcription engine to the latest version - Added automatic option to auto re-transcribe lines with repeated results after first time transcripe finished - Rewritten the segmented transcription feature: it now supports automatic segmentation based on pauses for long videos such as movies and animations, and fixes the previous issue of misaligned segment timing - Real-time transcription now supports Flash Attention for better performance - You can now customize the project name when creating a new real-time transcription project - Added hide translation engines not use - You can simply select one of the repeated sentences and click Re-transcribe; app will automatically select all repeated segments and open the transcription window - Redesigned and optimized the settings interface for real-time transcription - Fixed the jitter issue when auto-scrolling to the last line during real-time transcription - Fixed an issue where the translation module didn’t remember the last selected language - Fixed an issue where re-transcription failed after switching the engine for a selected segment - Other UI improvements and bug fixes

- The file saved after real-time transcription now directly displays the corresponding transcribed subtitles. - Subtitles in the ASS layout can be conveniently set to display in a top-and-bottom layout. - If the video file comes with text-based subtitle files, they will be automatically converted into project snapshots. - CJK Languages in Hardcoded videos now support controlling the maximum character limit per line. - Added a tool for batch hard line breaks in subtitle lines. - The translated text can be directly merged into the original transcription. (Subtitle editor context menu) - Fixed an issue where some files could not be accessed when attempting to read them. - Fixed a crash that occurred when stopping real-time transcription while recording through a microphone.

- Updated feedback feature (previous versions could not properly receive user feedback, please try contacting again) - Fixed several crash bugs - Fixed an issue causing crashes on macOS 13 - Fixed some crashes in the previewer - Fixed audio anomalies during real-time transcription and video recording

- Redesigned the logic for using the LLM plugin and added real-time debugging functionality for LLM effects - Added support for custom fonts in transcription and translation content - Optimized the real-time transcription parameters and file management interface - Exported files now support .mp4 format with subtitles - Supports directly pasting pyannote result text into the subtitle editor interface and parsing it - Supports using simple-one-api as a bridge for OpenAI format - During real-time transcription, identical transcription content will automatically be merged into one - Real-time transcription small window mode now supports reverse order display mode - Added multiple style switches for the real-time transcription small window - Fixed the issue where files could not be downloaded - Optimized global search functionality

- Add support whisperkit as new transcribe engine. (macOS 13+) - LLM execution supports operations on text in Text Mode - Quickly switch to text mode shortcut changed to ⌃+z - Fixed subtitle flickering in the previewer. - Fixed issues with range re-transcription.

- Add feedback feature to let users provide suggestions or detailed steps for issues. - Add auto open in Final Cut Pro after export .fcpxml format file - Add shortcut 'z' to quick switch between editor text view or table view - Fixed Real-time transcription float window press ESC will close window issue - Fixed export media range with real-time record file issue (.caf format) - Fixed url project can not access when move file to new location - Fixed an issue where certain formatted media files could not be exported. - Fixed an issue on macOS 12 and 13 could not start after updating to the latest version.